Appropriate comm_grps usage

GROMACS version: 2020.6
GROMACS modification: No

Hi all,

CHARMM-GUI has fixed their mdp file options in Membrane Builder to specify SOLU_MEMB and SOLV separately for comm_grps

I was just wondering about a special case:

I have a protein which is fully embedded in a membrane, and I have a protein which is partially solvated, but also partially bound to the cavity of the first protein near the membrane-solvent interface. It extends inside the first protein somewhat, but the majority of it is surrounded by solvent. What is the appropriate group for motion removal for the second protein? Should the second protein be part of SOLU_MEMB, or should I make a new group with SOLV_PROTEIN2?

Thanks for the assistance